Vince Analytics

Last reviewed Sep 13, 2026 · 732 words

Vince Analytics gives you a Plausible-shaped dashboard from one static binary and one data directory, with no ClickHouse, no PostgreSQL and no container orchestration. On a 1 GB VPS that already runs three other things, that is the entire pitch: Plausible Community Edition needs two databases and about 1 GB to itself, and Vince needs 128 MB and a mount point. The cost is a project born in 2023 with a small community, and you should weigh that against the convenience honestly rather than picking on resource use alone.

What a single binary changes operationally

Everything the dashboard shows lives in embedded storage under one directory. Backup is stopping the process or snapshotting the directory and copying it. Upgrade is replacing one file and restarting. There is no schema migration between two database engines to worry about, no ClickHouse memory tuning, and no cluster of containers where one restarts in a loop after a host reboot. For someone who wants numbers for a handful of personal sites, that operational shape is the reason to be here.

services:
  vince:
    image: ghcr.io/vinceanalytics/vince:latest
    ports:
      - "8080:8080"
    volumes:
      - ./vince-data:/data
    restart: unless-stopped

Put a reverse proxy with TLS in front, create the admin account on first visit, add a site by domain, and paste the snippet the site page hands you. The snippet is the Plausible pattern: a small deferred script with a data-domain attribute pointing at your own host. From there the dashboard is the one-page view Plausible made familiar: visitors, page views, bounce rate, visit duration, top pages, sources, countries, devices, and goals.

Cookieless counting, same rules as Plausible

Vince does not set cookies and does not store IP addresses; visitors are counted by a daily-rotating hash of the IP, user agent and site, so a returning visitor tomorrow is a new visitor tomorrow. That keeps it outside consent-banner territory in most readings of GDPR and ePrivacy, and it means "unique visitors" is a daily concept, not a lifetime one. Anyone coming from Google Analytics should read the switching guide before comparing numbers, because the two will not match and neither is wrong.

The ad-blocker problem applies in full. A first-party script on your own domain survives the generic filter lists better than a plausible.io script does, but a hostname or path containing "analytics" or "stats" still gets caught by some lists. Serve the script from a boring path on your main domain through the reverse proxy and expect to recover a portion, not all, of the blocked traffic.

What you give up against the bigger projects

Plausible has years of production hardening, a documented API, funnels, custom properties, a WordPress plugin ecosystem, and an obvious upgrade path to its paid cloud if self-hosting stops being fun. Umami runs on PostgreSQL only, supports many users and teams, and has the best multi-site management of the three. GoatCounter is Vince's closest relative in spirit, a single Go binary with SQLite, and has been around since 2019 with a track record to match. Vince's feature list sits between GoatCounter and Plausible, its community is the smallest of the four, and integrations that assume Plausible's API may or may not work against it. The Plausible versus Umami comparison covers the two safer picks if you want the long-run option.

Who actually runs it

A developer with 3 to 10 low-traffic sites on one small server, comfortable reading a GitHub repo when the docs run out, who wants Plausible's dashboard without Plausible's footprint. That person is well served. A business that needs analytics as a dependable, multi-user product with a support path is not the audience yet, and the project does not claim to be.

What I'd do

If resource use is the constraint, run Vince behind Caddy, snapshot the data directory nightly, and treat it as the analytics you can afford to lose and rebuild. If the constraint is confidence over the next 5 years, run Plausible CE and pay the 1 GB, or Umami if you have PostgreSQL already. And if you want the single-binary shape with more history behind it, GoatCounter is the same idea with 4 more years of bug fixes.